The biobank contains 256 organoids representing five cancer types with significant unmet medical needs: colorectal, oesophageal, pancreatic, stomach and ovarian cancers. Organoids are miniature three-dimensional cultures grown from patient tumour cells. Unlike conventional two-dimensional cell lines, in which cancer cells spread across the flat surface of a laboratory dish, organoids can reproduce important features of the original tumour’s architecture, genetic diversity and behaviour. They are not complete tumours, but they provide a more biologically realistic environment for studying how cancer cells grow and respond to treatment.

Creating the collection required a coordinated network of hospital and research centres in Birmingham, Cambridge, Glasgow, London and Southampton. Fresh tumour tissue donated by consenting patients was rapidly transported to the Sanger Institute, where researchers isolated viable cancer cells and placed them in carefully controlled culture conditions. These conditions included specialised growth factors and three-dimensional scaffolds or matrices that encourage cells to organise into structures resembling aspects of the tissue from which they originated. Establishing such cultures is technically demanding because tumour samples contain a mixture of cancer cells, immune cells, connective tissue and other normal cells, and not every sample forms a stable organoid.

To determine how faithfully the organoids represented the patients’ cancers, the researchers compared DNA sequences from the organoids with sequences from the original tumours and, where available, blood samples from the same patients. This approach allowed the team to distinguish inherited genetic variants from mutations acquired by the tumour and to monitor whether the models changed as they were maintained in the laboratory. The analysis showed that the organoids generally retained key genetic characteristics of the tumours from which they were derived, supporting their use as experimental models while also providing a way to track laboratory adaptation over time.

The researchers then applied CRISPR screening to 162 organoid models. CRISPR is a genome-editing technology that can be programmed to disrupt individual genes. In a screening experiment, thousands of cells receive different gene-targeting guides, and the population is monitored to determine which genetic disruptions prevent cells from surviving or multiplying. If cells carrying a particular guide disappear from the culture, the targeted gene may be essential under those conditions. By performing these screens across many tumour models, the team could distinguish broad cancer dependencies from vulnerabilities restricted to particular cancer types or molecular subgroups.

The screens identified thousands of dependencies. Some genes were required by many cancer models, reflecting fundamental processes such as DNA replication, protein production, cell division or energy metabolism. Others were important only in tumours carrying particular mutations or genomic changes. These selective dependencies are especially interesting for drug discovery because they may offer a route to target cancer cells while limiting damage to healthy tissues. However, a gene dependency observed in an organoid is not automatically a viable drug target; it must be validated through additional experiments, tested for safety and assessed in increasingly complex biological systems.

By integrating the CRISPR results with genomic and clinical information, the researchers identified 1,733 associations between gene dependencies and tumour features. These links included specific DNA alterations and treatment histories, helping to explain why genetically different cancers may respond differently to the same therapy. The resource also provided clues about how some tumours adapt after treatment. In several cases, organoids established from the same patient before and after therapy allowed the team to compare tumour states and identify changes associated with treatment resistance, as well as weaknesses that might be exploited by alternative approaches.

The findings do not represent an immediate new treatment for patients, and the organoids are not intended to replace clinical trials or traditional cancer models. Instead, the biobank is designed as an open research platform that can help scientists prioritise hypotheses before investing in drug development. Researchers can use the models to investigate cancer biology, test combinations of therapies, study resistance mechanisms and explore why a treatment succeeds in one molecular context but fails in another. Because the models are linked to patient data and genetic information, they may also help close the gap between laboratory discoveries and the biology of cancers observed in hospitals.

The study forms part of a broader international effort to improve next-generation cancer models. Two complementary papers published in Nature describe an expanded dependency map using genome-editing screens and the Human Cancer Models Initiative’s international collection of patient-derived models. Together, the studies point toward a more systematic era of cancer research in which experimental models are selected according to the genetic and clinical features they represent. Data from the new biobank will be made freely available through the Cell Model Passports website, while organoids are expected to be distributed through Merck and the nonprofit American Type Culture Collection, allowing laboratories worldwide to investigate cancer’s vulnerabilities with shared, better-characterised tools.

Subject of Research: Cancer gene dependencies, patient-derived tumour organoids and CRISPR screening

Article Title: A tumour-derived organoid biobank maps cancer gene dependencies

News Publication Date: 5 August 2026

Web References:
Wellcome Sanger Institute: https://www.sanger.ac.uk/
Cell Model Passports: https://cellmodelpassports.sanger.ac.uk/
Wellcome: https://wellcome.org/

References:
Herranz-Ors, C. et al. (2026), “A tumour-derived organoid biobank maps cancer gene dependencies,” Nature. DOI: 10.1038/s41586-026-10830-y
“A dependency map enhanced with next-generation 3D cancer models,” Nature. DOI: 10.1038/s41586-026-10843-7
“A compendium of next-generation patient-derived models for diverse cancers,” Nature. DOI: 10.1038/s41586-026-10806-y

Keywords: Cancer, oncology, tumour organoids, cancer biobank, CRISPR screening, gene dependencies, cancer vulnerabilities, precision medicine, treatment resistance, genomics, patient-derived models, drug discovery

The team behind this research has recognized a critical gap in the existing methodologies for predicting patient responses to TKIs. Traditionally, treatment decisions for NSCLC patients have relied heavily on genetic testing and basic clinical parameters; however, these approaches often lack the nuance and precision needed for effective treatment planning. By integrating multiple data modalities, including genomic, clinical, and imaging data, the researchers aim to provide a more holistic understanding of how patients with EGFR mutations will respond to TKI therapies.

An impressive array of data sources was harnessed for this study, including next-generation sequencing results, clinical trial databases, and advanced imaging techniques. By employing advanced machine learning algorithms, the authors were able to reveal patterns and correlations that have previously gone unnoticed within standard analytic frameworks. This cross-disciplinary approach has the potential to enhance not just treatment efficacy but also patient stratification, ensuring that individuals receive the most appropriate and effective therapies tailored uniquely to their tumor characteristics.

The importance of integrating these diverse data types cannot be overstated. In the context of advanced NSCLC, where tumor heterogeneity can greatly influence treatment outcomes, a multimodal approach allows for the nuanced understanding of how various factors interact to affect patient prognosis. This complexity has historically posed significant challenges in personalizing oncological care; however, the current study endeavors to dismantle these barriers and pave the way for more targeted therapeutic interventions.

Among the many findings presented in the study, the researchers discovered that specific genetic alterations within the EGFR gene could be more predictive of TKI therapy responses when analyzed in conjunction with imaging characteristics. This interplay between molecular and phenotypic data offers valuable insights into tumor behavior and can guide oncologists in selecting the most effective therapeutic regimens. Enhanced precision in prediction models not only helps in therapy selection but also in identifying patients who may benefit from alternative treatment modalities sooner.

Furthermore, the researchers employed rigorous validation processes to ensure the robustness and reliability of their predictive model. By utilizing datasets from several institutions around the globe, the authors were able to mitigate the risks of overfitting and bolster the model’s generalizability across diverse patient populations. This aspect of the study serves as a critical reminder of the importance of collaborative research in achieving statistically significant and clinically applicable findings.

The significance of their work extends beyond the immediate benefits to patient care; it also fosters a broader understanding of cancer biology and therapy response mechanisms. By elucidating the links between various data modalities and patient outcomes, the study contributes to the overall body of knowledge regarding precision medicine in oncology. This integrative approach may inspire future research initiatives aimed at identifying similar predictive markers in other cancer types.

As the study anticipates publication, the potentially transformative effects of its findings on clinical practice are already igniting discussions among oncologists and researchers alike. With the ever-evolving landscape of cancer treatments and the critical need for personalized approaches, the incorporation of robust predictive modeling could catalyze new standards of care in the near future.

What sets this research apart is not merely its innovative approach but also its timeliness. With the increasing approvals of novel TKI agents, understanding which patients will benefit most from these therapies is of utmost importance. As clinical trial landscapes become more crowded, effective patient selection strategies will be needed to navigate the complexities of modern cancer therapies successfully.

Patient empowerment is another crucial element addressed within the study. By producing predictive models that clinicians can rely upon, patients stand to benefit from informed discussions regarding their treatment options. Medical dialogues that prioritize patient involvement have the potential to enhance patient adherence and overall satisfaction with care.

While the study offers tremendous promise, it also raises important questions regarding future directions in cancer treatment research. How can similar methodologies be applied to other cancer types? Can the framework established by Chai and colleagues be adapted for a broader array of therapeutics beyond TKIs? These inquiries highlight the study’s role as a launching pad for continued exploration in the field.

In conclusion, the multidisciplinary research conducted by Chai, Li, Yang, and their team marks a significant milestone in the quest for personalized oncology. By integrating multifaceted data sources to predict TKI outcomes in advanced EGFR-mutated NSCLC patients, this work stands to change the standard of care for many individuals suffering from this debilitating disease. As the medical community eagerly awaits the full publication and implications of these findings, it is clear that the future of lung cancer treatment may be brighter than ever before.

Subject of Research: Multimodal prediction of tyrosine kinase inhibitors therapy outcomes in advanced EGFR-mutated NSCLC patients

Article Title: Multimodal prediction of tyrosine kinase inhibitors therapy outcomes in advanced EGFR-mutated NSCLC patients

Article References: Chai, X., Li, H., Yang, M. et al. Multimodal prediction of tyrosine kinase inhibitors therapy outcomes in advanced EGFR-mutated NSCLC patients. J Transl Med 23, 933 (2025). https://doi.org/10.1186/s12967-025-06956-8

Image Credits: AI Generated

DOI: 10.1186/s12967-025-06956-8

Keywords: Tyrosine Kinase Inhibitors, EGFR-mutated NSCLC, Multimodal Prediction, Personalized Medicine, Advanced Cancer Therapies, Machine Learning.